Re: Linux 2.4.18-pre9-ac1

[Jens Axboe <axboe@suse.de>]

On Tue, Feb 12 2002, Jens Axboe wrote:
> On Tue, Feb 12 2002, Alan Cox wrote:
> > > I notice that in linux/drivers/scsi/scsi_merge.c you seem to
> > > be reverting the MO drive clustering fix from Jens:
> > >     http://www.uwsg.indiana.edu/hypermail/linux/kernel/0202.0/1321.html
> > > 
> > > Was this intentional?  If so, why?
> > 
> > I want to find out why it was done first and then test it. Leaving it out
> > will ensure it bugs me until I test it
> 
> If you leave it out, you surely want to make sure that the other request
> init and re-init paths agree on the clustering for MO devices. Because
> they don't.
> 
> As far as I'm concerned, removing the MO conditional wrt clustering is
> the right fix.

BTW, if you are concerned with the write/read vs seek latencies of MO
drives, then the disable clustering hack was definitely the wrong way to
try and limit request sizes. In fact it achieved absolutely _nothing_.
Clustering at this level is completely device independent, too.

In short, the old code made no sense whatsoever.

Now, disabling request merging for MO devices might make a whole lot
more sense. That might be worth while trying, and I'd be happy to give
you a patch to try that out instead.

-- 
Jens Axboe